|Home | About | Journals | Submit | Contact Us | Français|
Despite widespread expression of EGF receptors (EGFRs) and EGF family ligands in non-small cell lung cancer (NSCLC), EGFR-specific tyrosine kinase inhibitors (TKIs) such as gefitinib exhibit limited activity in this cancer. We propose that autocrine growth signaling pathways distinct from EGFR are active in NSCLC cells. To this end, gene expression profiling revealed frequent co-expression of specific fibroblast growth factors (FGFs) and FGF receptors (FGFRs) in NSCLC cell lines. Notably, FGF2 and FGF9 as well as FGFR1 IIIc and/or FGFR2 IIIc mRNA and protein are frequently co-expressed in NSCLC cell lines, especially those that are insensitive to gefitinib. Specific silencing of FGF2 reduced anchorage-independent growth of two independent NSCLC cell lines that secrete FGF2 and co-express FGFR1 IIIc and/or FGFR2 IIIc. Moreover, a TKI (RO4383596) that targets FGFRs inhibited basal FRS2 and ERK phosphorylation, two measures of FGFR activity, as well as proliferation and anchorage-independent growth of NSCLC cell lines that co-express FGF2 or FGF9 and FGFRs. By contrast, RO4383596 influenced neither signal transduction nor growth of NSCLC cell lines lacking FGF2, FGF9, FGFR1 or FGFR2 expression. Thus, FGF2, FGF9 and their respective high-affinity FGFRs comprise a growth factor autocrine loop that is active in a subset of gefitinib-insensitive NSCLC cell lines.
Autocrine growth factor production by cancer cells provides self-sufficiency in growth signals, one of the six hallmarks of cancer (Hanahan and Weinberg, 2000). Based on the frequent expression of EGFR in NSCLC (Hirsch et al., 2003) as well as the widespread co-expression with various EGF family ligands (Rusch et al., 1997), the EGFR is an attractive candidate for a receptor tyrosine kinase (RTK) mediating autocrine growth in NSCLC. In this context, the EGFR inhibitors, gefitinib and erlotinib, were deployed in a series of clinical trials, but yielded response rates of only 10 to 20% (Dancey, 2004; Hirsch and Bunn, 2005). Subsequent molecular analysis of the responsive lung tumors revealed a significant enrichment for gain-of-function EGFR mutations (Han et al., 2005; Lynch et al., 2004). The general insensitivity to EGFR-specific TKIs is also reflected in cultured NSCLC cell lines (Coldren et al., 2006; Helfrich et al., 2006). Thus, despite broad expression of EGFR in NSCLC, only a subset responds to EGFR inhibitors. While rare mutations in EGFR that render the tyrosine kinase resistant to gefitinib and erlotinib have been identified (Pao et al., 2005), the limited response of NSCLC to EGFR inhibitors may reflect the activity of additional receptor tyrosine kinase systems distinct from EGFR (Morgillo and Lee, 2005; Rikova et al., 2007). Candidate receptor tyrosine kinase systems include cMet (Engelman et al., 2007; Lutterbach et al., 2007), Axl (Shieh et al., 2005; Wimmel et al., 2001) and IGF-1R (Morgillo and Lee, 2005) as well as novel tyrosine kinases such as Alk and Ros (Rikova et al., 2007). Moreover, a recent study by Stommel et al (Stommel et al., 2007) in glioblastoma cell lines provides strong evidence for the coactivation of multiple RTKs in cancer cells and the requirement for simultaneous blockade to achieve significant growth inhibition.
We examined archived gene expression array data obtained from a panel of NSCLC cell lines (Coldren et al., 2006) and noted frequent co-expression of distinct FGFs and FGFRs, suggesting that an FGFR-dependent autocrine signaling pathway may operate in a significant fraction of NSCLC. FGFs have been discovered from nematodes to man with 22 distinct FGFs presently identified in mammals (reviewed in (Eswarakumar et al., 2005; Grose and Dickson, 2005; Mohammadi et al., 2005)). FGFs stimulate diverse responses in development and tissue maintenance by binding to and activating a family of four receptor tyrosine kinases designated FGFR1 - FGFR4. The extracellular domain of FGFRs contains two or three immunoglobulin-like (Ig-like) loops where the two membrane-proximal loops encode the FGF binding site. Of particular importance to FGF binding specificity is the third Ig loop, the N-terminal half of which is encoded by an invariant IIIa exon with alternative usage of IIIb or IIIc exons for the C-terminal half (Eswarakumar et al., 2005; Mohammadi et al., 2005). As a general rule, FGFRs encoding exon IIIb (FGFR IIIb) are expressed on epithelial cells while the FGFRs encoding exon IIIc (FGFR IIIc) are expressed on mesenchymal cells (Mohammadi et al., 2005). By contrast, the ligands for FGFR IIIb are often expressed in mesenchymal cells while ligands for FGFR IIIc are expressed in epithelial cells. This establishes a paracrine mechanism of signaling between epithelia and mesenchyme that is critical to normal development and tissue homeostasis. In this regard, FGF7 and FGF10 expressed in lung mesenchymal compartments exhibit high affinity for FGFR2 IIIb expressed on lung epithelial cells, providing key signals required for epithelial-mesenchymal interactions that are pivotal to embryonic lung development. By contrast, FGF2 and FGF9 exhibit negligible binding to FGFR2 IIIb, but high affinity binding to FGFR1 IIIc and FGFR2 IIIc.
The literature provides ample precedent for the involvement of distinct FGFs and FGFRs in cancers of epithelial origin (reviewed in (Grose and Dickson, 2005)) including prostate, thyroid, skin, head and neck and urinary bladder. In many instances, tumorigenesis is associated with amplified FGFR expression, although somatic mutations within the receptor coding sequences that confer gain-of-function have also been identified. In addition to FGFR mutations, inappropriate expression of FGF ligands presents an alternative mechanism by which FGFRs could be activated and participate in oncogenesis. Our present study demonstrates co-expression of FGF2 or FGF9 as well as FGFR1 and/or FGFR2 in multiple NSCLC cell lines. Moreover, studies employing molecular silencing of FGF2 and a pharmacological FGFR inhibitor reveal a functional role for an FGFR autocrine signaling pathway in NSCLC cell lines, especially those that are insensitive to EGFR-specific TKIs.
The NSCLC cell lines employed in this study were submitted to fingerprint analysis by the University of Colorado Cancer Center to verify their authenticity. All cell lines were routinely cultured in RPMI-1640 growth medium supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) at 37°C in a humidified 5% CO2 incubator. Where indicated, the cells were switched to HITES medium (RPMI-1640 containing 10 nM hydrocortisone, 5 μg/ml insulin, 10 μg/ml transferrin, 10 nM estradiol, 30 nM Na3SeO3 and 1% bovine serum albumin) to limit mitogenic inputs from serum components.
Total RNA (3 μg) was reverse transcribed in a volume of 20 μl using random hexamers and MMLV reverse transcriptase. Aliquots (1 μl) of 10-fold diluted reverse transcription reactions were subjected to PCR in 25 μl reactions with SYBR® green Jumpstart Taq Readymix (Sigma, St. Louis, MO) and the primers listed in Table 1 using an I Cycler (BioRad, Hercules, CA). The real-time PCR amplification products from initial experiments were resolved by electrophoresis on 5% polyacrylamide gels to verify that the primer pairs amplified a single product of the predicted size. GAPDH mRNA levels were measured by quantitative RT-PCR in replicate samples as a housekeeping gene for normalization of the different mRNA expression and the data are presented as “Relative Expression”.
To measure the effect of FGFR inhibition on cell proliferation, NSCLC cells were seeded in 96-well plates at 1,000 cells/well, RO4383596 was added the next day at concentrations ranging from 0 to 3 μM and cultured for 5 additional days. Viable cells were subsequently estimated with a modified MTT assay as previously described (Carmichael, 1988). For measurement of anchorage-independent cell growth, 20,000 cells were suspended in 0.75 ml HITES containing 5% fetal bovine serum and 0.4% Difco™ agar noble (Becton, Dickinson and Co., Sparks, MD) and overlayed on base layers containing 0.75 ml HITES containing 5% fetal bovine serum and 0.5% agar noble in 12-well plates. In experiments testing the effect of RO4383596 on anchorage-independent growth, the drug was added to the base layers at two-fold concentrations such that diffusion into the top layers yielded the desired final drug concentrations. The plates were incubated in a 37°C CO2 incubator for 21 days and viable colonies were stained for 24 hrs with nitroblue tetrazolium. Following digital photography, the colonies were quantified using the MetaMorph imaging software program (Molecular Devices, Downingtown, PA).
NSCLC cell lines seeded in 12-wells plates were cultured in 1 ml HITES medium for three days. Subsequently, the medium was collected and assayed for FGF2 using a Quantikine- human FGF basic assay kit (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. The cell monolayers were lysed in 250μl MAP kinase lysis buffer (MKLB; 0.5% Triton X-100, 50 mM β-glycerophosphate (pH 7.2), 0.1 mM Na3VO4, 2 mM MgCl2, 1 mM EGTA, 1mM DTT, 0.3 M NaCl, 2 μg/ml leupeptin and 4 μg/ml aprotinin) and cell protein was measured by the Bradford assay. Secreted FGF2 measured by ELISA was normalized to cellular protein and presented as pg FGF2/mg cell protein.
Two distinct shRNAs (clone ID TRCN0000003329 targeting FGF2 sequences 799-819 and clone ID TRCN0000003332 targeting FGF2 sequences 756-776) encoded in the pLKO.1 lentiviral vector were obtained from Open Biosystems (Huntsville, AL). The pLKO.1 constructs encoding the FGF2 shRNAs as well as pLKO.1 encoding a control shRNA targeting GFP were packaged into lentiviruses and used to transduce H226 and H1703 cells according to the manufacturer’s protocol. Stable transfectants were selected with 1 μg/ml puromycin and drug-resistant cells were pooled and screened for FGF2 secretion by ELISA. The cultures were submitted to anchorage-independent growth assays within 3 passages following the initial puromycin selection.
For analysis of phospho-ERK and phospho-FRS2, growth factor or drug-treated NSCLC cells were rinsed once with phosphate-buffered saline, lysed in MKLB and centrifuged (5 min at 13,000 RPM). The particulate fractions were discarded and the soluble extracts were mixed with SDS sample buffer and submitted to SDS-PAGE. Following electrophoretic transfer onto nitrocellulose, the filters were blocked in 3% bovine serum albumin (Cohn Fraction V, ICN Biomedicals, Inc., Aurora, OH) in Tris-buffered saline with 0.1% Tween 20 (TTBS) and then incubated with rabbit polyclonal anti-phospho-ERK or rabbit polyclonal phospho-FRS2-Y196 antibodies (Cell Signaling Technology, Inc, Danvers, MA) for 16 hours at 4°C. The filters were washed thoroughly in TTBS, then incubated with alkaline phosphatase-coupled goat anti-rabbit antibodies and developed with LumiPhos reagent (Pierce, Rockford, IL) according to the manufacturer’s instructions. The filters were subsequently stripped and reprobed for total ERK1 and ERK2 using a mixture of rabbit polyclonal anti-ERK1 (sc-93) and ERK2 (sc-154) antibodies or FRS2 (sc-8318) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Because ERK2 was more abundant relative to ERK1 in the NSCLC cell lines, the respective phospho-ERK2 and total ERK2 bands were submitted to densitometry analysis and the data are presented as the ratio of phospho-ERK2 to total ERK2.
For immunoblot analysis of FGFR1, FGFR2, EGFR and the α-subunit of NaK-ATPase, NSCLC cells were collected in phosphate-buffered saline, centrifuged (5 min, 1000 × g) and suspended in hypotonic lysis buffer (10 mM Tris-Cl (pH 8.0), 1 mM EDTA, 4 μg/ml PMSF, 2 μg/ml leupeptin, 4 μg/ml aprotinin, 1mM DTT). The cells were homogenized by 6 to 8 passes through a 26 gauge syringe needle and the homogenate was submitted to centrifugation at 1000 × g for 5 minutes to pellet nuclei and unbroken cells. The resulting supernatants were centrifuged (10,000 × g for 10 min) to collect the membrane fragments that were subsequently resuspended in hypotonic lysis buffer. Aliquots of the membrane preparations containing 100 μg protein were submitted to SDS-PAGE and immunoblotted for FGFR1 (sc-57132), FGFR2 (sc-122) and NaK-ATPase α-subunit (sc-21712) with antibodies from Santa Cruz Biotechnology (Santa Cruz, CA). EGFR was detected with a rabbit polyclonal antibody (#2232) from Cell Signaling Technology, Inc.
When a dataset from the recent gene expression profiling of a panel of NSCLC cell lines (Coldren et al., 2006) was queried for the co-expression of known receptor tyrosine kinases and their respective polypeptide growth factors, frequent expression of FGF2, FGF9, FGFR1 and FGFR2 was observed (data not shown). The PDGF and VEGF family of growth factors were also predicted to be widely expressed in NSCLC cell lines, but the expression values for PDGFRs (PDGFRα, PDGFRβ) and VEGFRs (FLT1, FLK1) were uniformly low or absent with the exception of H1703 cells which express PDGFRα (Rikova et al., 2007). Quantitative RT-PCR analyses of 33 NSCLC cell lines confirmed the gene expression profiling data and revealed a non-Gaussian distribution of FGF2 and FGF9 mRNA expression (Figure 1A). By contrast, only one NSCLC cell line expressed FGF7 at a level different from the low to undetectable expression observed in the other NSCLC cell lines. Moreover, FGFR1 and FGFR2 mRNAs were also widely expressed in a non-Gaussian manner among the 33 NSCLC cell lines (Figure 1B). By contrast, FGFR3 and FGFR4 mRNAs were more restricted (not shown) and did not correlate with FGF2 or FGF9 expression or response to an FGFR TKI described below. Based on these findings, we highlighted FGF2, FGF9, FGFR1 and FGFR2 as putative components of an autocrine signaling pathway in NSCLC cell lines for further exploration.
A panel of NSCLC cell lines that had previously been characterized for sensitivity to the EGFR inhibitor, gefitinib (Coldren et al., 2006; Helfrich et al., 2006), was assayed for FGF2 and FGF9 mRNA using quantitative RT-PCR with primers listed in Table 1. NSCLC cell lines H226, H520, H661, H1299, H1703 and Colo699 are insensitive to gefitinib (IC50 > 4 μM) while H322c, H358, H1648, HCC827 and HCC4006 are sensitive to gefitinib (IC50 < 0.5 μM). Among the latter, HCC827 and HCC4006 bear EGFR exon 19 deletions conferring gain-of-function. As shown in Figure 2A, high FGF2 and FGF9 mRNA expression largely segregated with the gefitinib-insensitive NSCLC cell lines, although HCC827 and HCC4006 expressed detectable FGF2 mRNA. The mean relative FGF2 mRNA expression in the group comprised of H226, H520, H661, H1299, H1703 and Colo699 cells (0.88 ± 0.29) was statistically greater (p<0.02) than for H322c, H358, H1648, HCC827 and HCC4006 cells (0.092 ± 0.057). Moreover, analysis of conditioned medium from the NSCLC cell lines by ELISA revealed that FGF2 protein was readily secreted into the medium by NSCLC cells that expressed significant FGF2 mRNA (Fig. 2B).
Similar analysis of FGFR1 and FGFR2 mRNA levels in the NSCLC cell lines revealed exclusive segregation of FGFR1 mRNA in the cell lines previously defined as being gefitinib-insensitive (Fig. 3A). The mean FGFR1 mRNA relative expression in the group comprised of H226, H520, H661, H1299, H1703 and Colo699 cells (0.96 ± 0.43) was statistically greater (p<0.005) than the mean expression for H322c, H358, H1648, HCC827 and HCC4006 cells (0.006 ± 0.002). By contrast, the expression of FGFR2 mRNA was not different between the two groups of cells (Fig. 3A). FGFR1 protein as assessed by immunoblot analysis coincided closely with FGFR1 mRNA measured by quantitative RT-PCR (Fig. 3B). While FGFR2 protein was readily detected in H1648 and Colo699 cells that express high levels of FGFR2 mRNA (Fig. 3A), a significant level of FGFR2 protein was also detected in H226, H520 and H661 cells which expressed very low levels of FGFR2 mRNA. Thus, FGF2 or FGF9 and FGFR1 and/or FGFR2 are co-expressed in H226, H520, H661, H1299, H1703 and Colo699 cells. By contrast, HCC827 and HCC4006 express FGF2, but no detectable expression of FGFR1 or FGFR2 mRNA or protein.
FGFR1 and FGFR2 mRNAs are alternatively-spliced within the third extracellular, membrane-proximal immunoglobulin-like loop to generate the respective FGFR IIIb and FGFR IIIc proteins which display distinct FGF family member binding properties (Eswarakumar et al., 2005; Grose and Dickson, 2005; Mohammadi et al., 2005). FGFR1 IIIb, FGFR1 IIIc and FGFR2 IIIc bind FGF2 and FGF9 with high affinity while FGFR2 IIIb binds FGF7 and FGF10. We employed published (Kwabi-Addo et al., 2001; Murgue et al., 1994) PCR strategies to amplify the regions of the FGFR1 and FGFR2 cDNAs flanking the third extracellular immunoglobulin loops and the resulting cDNA products were digested with restriction enzymes that discriminate between the FGFR IIIb and FGFR IIIc sequences. As shown in Figure 4, the FGFR1 PCR products amplified from cDNA prepared from Colo699, H520, H1703, H1299 and H226 cells were completely digested with AflII which cuts within the FGFR1 IIIc sequence, but not Bsm1, which cuts the product arising from the FGFR1 IIIb mRNA. Analysis of FGFR1 mRNA from H661 cells also revealed only FGFR1 IIIc (not shown). Thus, FGFR1 IIIc is the sole FGFR1 isoform expressed in NSCLC cells. Similar analysis of FGFR2 PCR products revealed expression of FGFR2 IIIc (cut by Hinc II) in Colo699 and H226 cells (Fig. 4) as well as H520 and H661 (not shown). By contrast, H322c, H358, H1648, HCC827 (Fig. 4) and HCC4006 (not shown) exclusively expressed the FGFR2 IIIb splice product that is digested by Ava I (Fig. 4). Therefore, expression of FGF2 and/or FGF9 in H226, H520, H661, H1299, H1703 and Colo699 cells coincides with expression of one or more of their high-affinity receptors, FGFR1 IIIc and/or FGFR2 IIIc. While FGF2 mRNA and protein are expressed in HCC827 and HCC4006 cells, neither FGFR1 IIIc nor FGFR2 IIIc were expressed in these cell lines.
To determine if the co-expression of FGFs and FGFRs in NSCLC cell lines is also observed in primary NSCLC tumors, the expression of FGF2, FGF9, FGFR1 and FGFR2 mRNAs was measured by quantitative RT-PCR in total RNA prepared from nine primary human NSCLC tumor samples obtained through the University of Colorado Lung SPORE. The findings in Table 2 show expression values for FGF2 and FGF9 where 6 of the 9 tumors examined expressed FGF2 or FGF9 at levels greater than median values (0.13 and 0.03 for FGF2 and FGF9, respectively). In addition, FGFR1 mRNA was expressed at levels similar to those detected in NSCLC cell lines and was exclusively spliced to yield FGFR1 IIIc (not shown). Likewise, FGFR2 was expressed at levels greater than the median level in 4 of the 9 tumors and analysis of the IIIb/IIIc status of FGFR2 revealed FGFR2 IIIc expression, alone or in combination with FGFR2 IIIb, in 4 of the 9 tumors. Thus, these data indicate that FGF2 and FGF9 are frequently co-expressed with FGFR1 IIIc and/or FGFR2 IIIc in primary NSCLC tumors.
We tested whether NSCLC cell lines expressing FGFR1 IIIc or FGFR2 IIIc responded to exogenous FGF2 as predicted by their FGFR isoform status. The NSCLC cell lines were incubated with EGF or FGF2 for 15 minutes and cell extracts were submitted to immunoblot analysis of phospho-ERK as a measure of receptor activation. As shown in Figure 5, FGF2-treatment of NSCLC cell lines expressing FGFR1 IIIc and/or FGFR2 IIIc (H226, H520, H661, H1299, H1703, Colo699) stimulated ERK phosphorylation. H520, H661 and Colo699 cells failed to respond to EGF, consistent with little or no expression of EGFR in these lines (see Fig. 3B). By contrast, H322c, H358, H1648, HCC827 and HCC4006 responded to EGF, but not exogenous FGF2, with increased phospho-ERK. The NSCLC cell lines exhibit the FGF2-stimulated phospho-ERK responses predicted by the FGFR isoform expression status.
Combined, the findings in Figures Figures22--55 demonstrate co-expression of FGF2 or FGF9 and functional FGFR1 or FGFR2 in H226, H520, H661, H1299, H1703 and Colo699 cells. To directly test the role of FGF2 as an autocrine growth factor in H226 and H1703 cells, endogenous FGF2 mRNA was silenced by stable expression of shRNA constructs (see Materials and Methods). Compared to a control shRNA targeting GFP, the FGF2-specific shRNAs (3329 and 3332) mediated a significant reduction in FGF2 mRNA (not shown) and FGF2 secretion (Figure 6A). Also, silencing of FGF2 in H226 and H1703 cells decreased basal phospho-ERK levels relative to cells transduced with the GFP control (Figure 6B), indicating decreased proximal activity of FGFRs. Interestingly, the decrease in pERK levels in H1703 cells expressing the FGF2 shRNA was not as extensive as observed in H226 cells, possibly resulting from the ongoing activity of the activated PDGFRα present in H1703 cells (Rikova et al., 2007). Finally, anchorage-independent growth of both H226 and H1703 cells transduced with the FGF2 shRNAs were inhibited compared to transfectants expressing the control GFP shRNA (Fig. 6C). In H226 cells, the reduction in soft agar growth by the two distinct shRNAs was consistent with the degree of reduction in FGF2 expression. These data provide molecular evidence for the role of FGF2 as an autocrine factor in H226 and H1703 cells.
RO4383596 is a small molecule inhibitor of the related family of FGFRs, VEGFRs and PDGFRs (McDermott et al., 2005). Published studies demonstrated that RO4383596 directly inhibits VEGFR, PDGFR and FGFRs with an IC50 ~30-40 nM, blocks autophosphorylation of VEGFR stimulated by exogenous VEGF and inhibits growth factor-stimulation of effector signal pathways including ERK1/2 and Akt (McDermott et al., 2005). Importantly, RO4383596 inhibited both VEGF and FGF2-stimulated proliferation of human umbilical cord endothelial cells (McDermott et al., 2005). The Affymetrix data set (Coldren et al., 2006) reveals little or no expression of PDGFRα and PDGFRβ mRNA in NSCLC cell lines with the exception of H1703 cells which express PDGFRα (not shown). In addition, mRNA levels for the VEGFRs, FLT1 and FLK1, were less than 2% and 0.4%, respectively, of the levels measured in human microvascular endothelial cells as assessed by quantitative RT-PCR (data not shown). Thus, RO4383596 was employed as a TKI to test the requirement of the putative FGF-FGFR autocrine pathway for transformed growth of NSCLC cells with the caveat that H1703 cells express an additional target, PDGFRα, for this drug.
The ERK MAP kinases represent a conserved signal pathway activated distal to diverse growth factor receptors as well as oncogenes such as Ras and Src. By contrast, the FGF receptor substrate substrate 2 (FRS2) docking protein serves as a proximal component of FGFR and Trk-mediated signal pathways (Eswarakumar et al., 2005). Treatment of H226 cells with FGF2 increased both ERK and FRS2-Y196 phosphorylation (Fig. 7) and both activities were inhibited in a dose-dependent manner by RO4383596. EGF treatment elicited a small effect on FRS2 phosphorylation and strongly increased ERK phosphorylation. Neither response was inhibited by RO4383596, indicating that neither the EGFR nor conserved signaling components that participate in growth factor-stimulated ERK phosphorylation are targets for RO4383596.
When the basal phosphorylation states of ERK1 and ERK2 are used as a measure of growth factor and oncogene signaling in NSCLC cell lines cultured in serum-free HITES medium without exogenous growth factors, RO4383596 elicited a dose-dependent inhibition of phospho-ERK in H226, H520, H661, H1299, H1703 and Colo699 cells (Fig. 8A and B). By contrast, little or no influence of RO4383596 on pERK levels in H322c, H358, H1648, HCC827 and HCC4006 cells was observed (Fig. 8A and B). Similar treatment of the different NSCLC cell lines with gefitinib revealed no inhibition of pERK levels in H226, H520, H661, H1299, H1703 or Colo699 cells, but strong inhibition in H322c, H358, H1648, HCC827 and HCC4006 cells (Fig. 8A and B), confirming their previously-described sensitivity to EGFR-specific TKIs (Coldren et al., 2006). As a proximal measure of FGFR signaling status in the NSCLC cell lines, the phosphorylation status of FRS2 was monitored in several NSCLC cell lines treated with or without RO4383596. As shown in Figure 8C, the intensity and mobility of pFRS2 was markedly altered upon RO4383596-treatment in H226, H520 and Colo699 cells, but not in H322c, H358 and H1648, indicating that basal FRS2 phosphorylation is regulated by FGFRs in the former cell lines. Combined, these findings indicate that an RO4383596-sensitive FGFR-mediated autocrine signaling pathway contributes to the phosphorylation and activity of FRS2 and ERK in H226, H520, H661, H1299, H1703 and Colo699 cells while, as predicted, an EGFR-dependent pathway is dominant for maintaining the phosphorylation of ERK1 and ERK2 in H322c, H358, H1648, HCC827 and HCC4006 cells.
Analysis of the influence of RO4383596 on proliferation and anchorage-independent growth of the NSCLC cell lines revealed dose-dependent inhibition of proliferation (Fig. 9A) and/or anchorage-independent growth (Fig. 9B) of H226, H520, H661, H1299, H1703 and Colo699 cells by RO4383596. By contrast, proliferation or anchorage-independent growth of gefitinib-sensitive H322c, H358, H1648, HCC827 and HCC4006 cells was not significantly influenced by RO4383596. Thus, RO4383596 inhibits ERK pathway activation and proliferation/anchorage-independent growth in a set of gefitinib-insensitive NSCLC cell lines that co-express FGF2 or FGF9 and FGFR1 IIIc or FGFR2 IIIc and supports the involvement of FGF and FGFR autocrine signaling as a dominant pathway mediating transformed growth in a subset of NSCLC cells. It is not possible to assign an exclusive autocrine role for FGF2 and FGFR1 in H1703 cells from their sensitivity to RO4383596. As previously mentioned, these cells also express activated PDGFRα (Rikova et al., 2007) and RO4383596 is an effective inhibitor of this receptor tyrosine kinase as well (McDermott et al., 2005). Based on the ability of FGF2 silencing to reduce ERK activity and anchorage-independent growth, it is likely that both FGFR1 and PDGFRα contribute to transformed growth in H1703 cells.
Previously published studies have demonstrated expression of FGF2 and FGFRs in human lung cancers and in NSCLC cell lines (Berger et al., 1999; Chandler et al., 1999; Kuhn et al., 2004). Also, rare somatic mutations in FGFR1 that may confer gain-of-function have been identified (Zhao et al., 2005) and amplification of the FGFR1 gene has been detected in human NSCLC, albeit at a very low frequency (Davies et al., 2005). Our results support these previous studies and also provide molecular evidence for an active FGF autocrine signaling pathway in a subset of NSCLC cell lines. Our demonstration of an active FGF-FGFR autocrine loop in NSCLC cell lines also provides a mechanism for the observed insensitivity of some NSCLC tumors and cell lines to EGFR-specific TKIs. Our data suggest that gefitinib-insensitive NSCLC cell lines employ alternative receptor tyrosine kinases, such as the FGFR, to establish self-sufficiency in growth. Previous studies have shown that gefitinib-sensitive NSCLC tumors and cell lines are enriched for the adenocarcinoma and bronchoalveolar carcinoma histological type (Miller et al., 2004). By contrast, among the NSCLC cell lines where we demonstrated FGFR-dependent autocrine signaling, H226, H520 and H1703 are derived from squamous cell carcinomas and H661 and H1299 are derived from large cell carcinomas. Thus, FGFR inhibitors may better target the squamous cell and large cell histologies of NSCLC cell lines that frequently exhibit a high degree of insensitivity to EGFR TKIs.
A recent study (Rikova et al., 2007) employed a phosphotyrosine proteomic approach to survey for tyrosine kinases that are active in lung cancer cell lines and primary tumors. In contrast to our present findings, only 3 of 41 NSCLC cell lines showed evidence for activated FGFR1, suggesting that the FGF/FGFR pathway is not a major receptor pathway active in NSCLC cells lines. It is noteworthy that the NSCLC cell line panel employed in the aforementioned study was comprised of ~75% adenocarcinomas while we find that squamous cell and large cell carcinomas are enriched in the FGF and FGFR-expressing NSCLC cell lines. In addition, each experimental approach (proteomics, genomics, etc) will have unique biases and tyrosine-phosphorylated peptides derived from FGFR1 or FGFR2 may not be efficiently detected by the experimental approaches employed. Alternatively, the selected phosphopeptides from FGFR1 or FGFR2 may not serve as sensitive indicators of FGFR activity in cell lines or tumors.
A conclusion from our experiments is that FGFR1 IIIc and/or FGFR2 IIIc and their respective ligands, FGF2 and FGF9, undergo co-selection as components of an autocrine signaling pathway during initiation and progression of NSCLC. It is noteworthy that no NSCLC cell line expressing FGFR1 IIIc or FGFR2 IIIc was found that did not also express FGF2 or FGF9. However, HCC827 and HCC4006 expressed FGF2, but lacked detectable expression of FGFR1 IIIc or FGFR2 IIIc. As single markers, FGFR1 IIIc or FGFR2 IIIc may be a more reliable indicator of FGF and FGFR-dependent autocrine signaling. FGFR signaling is critical for lung development and tissue homeostasis, although FGFR2 IIIb, FGF7 and FGF10 are the key players in this regard (Eswarakumar et al., 2005). However, our studies did not detect co-expression of FGFR2 IIIb with FGF7 or FGF10 in human NSCLC cell lines. Rather, FGFR1 IIIc and FGFR2 IIIc are co-expressed with their high-affinity ligands, FGF2 or FGF9. The literature indicates that FGF2 and FGFR1 appear to be most highly expressed in vascular compartments of the lung (Powell et al., 1998). Thus, ectopic expression of FGF2, FGF9, FGFR1 IIIc and FGFR2 IIIc may be induced during lung tumorigenesis. In support of this possibility, enhanced bronchial expression of FGF2 and FGFR1 is observed in chronic obstructive pulmonary disease (Kranenburg et al., 2002; Kranenburg et al., 2005) where the risks for chronic obstructive pulmonary disease and lung cancer aggregate (Schwartz and Ruckdeschel, 2006). An alternative mechanism for the observed co-expression of specific FGFs and FGFRs in NSCLC cell lines is that a specific lung epithelial cell type or progenitor may express FGF2 or FGF9 and FGFR1 IIIc and/or FGFR2 IIIc and serve as a precursor for the NSCLC cells that exhibit an FGFR autocrine signaling loop. In support, a recent study by Ince et al revealed a higher degree of gene expression similarity between a given non-transformed mammary epithelial cell type and its transformed derivative than between different types of non-transformed mammary epithelial cell precursors (Ince et al., 2007), indicating that tumor cells retain a high degree of gene expression similarity with their non-transformed precursors.
In addition to autocrine signaling through EGFR and FGFR pathways in NSCLC, the literature documents that multiple receptor tyrosine kinases will participate in lung oncogenesis (Rikova et al., 2007) including cMet, which is amplified in some gefitinib-insensitive NSCLC (Engelman et al., 2007; Lutterbach et al., 2007). Inspection of the gene expression array dataset derived from the NSCLC cell lines (Coldren et al., 2006) also predicts the co-expression of the Axl family of receptor tyrosine kinases and their ligands, Gas6 and protein S (Hafizi and Dahlback, 2006), in a significant fraction of NSCLC cell lines. In fact, a potential role for an Axl/Gas6 autocrine loop in NSCLC has been previously invoked (Shieh et al., 2005; Wimmel et al., 2001). Finally, an IGF-1R signaling system has been proposed as a transforming pathway in NSCLC (Morgillo and Lee, 2005). Our own preliminary studies reveal frequent over-expression of the IGF-1 receptor in NSCLC cell lines, although co-expression of the ligands, IGF1 or IGF2 is less clear. However, it is important to consider paracrine involvement of growth factor receptors as well where the tumor microenvironment may provide growth factors that stimulate receptor tyrosine kinases expressed on lung cancer cells. Combined, our present study and the literature indicate that the EGFR does not function as a single dominant receptor tyrosine kinase in autocrine growth of NSCLC, but that multiple autocrine loops will participate. Thus, effective blockade of autocrine and paracrine signaling in primary NSCLC tumors will require precise identification of the active receptor tyrosine kinase pathways through appropriate biomarkers. The fact that growth factors such as FGF2 are secreted molecules appearing in extracellular fluids of cancer patients (Nguyen et al., 1994) suggests that the autocrine factors, themselves, may serve as informative biomarkers in this regard.
The studies were supported by NIH grants R01 CA116527, R01 CA127105, P30 CA046934 and P50 CA58187 and a Cancer League of Colorado grant to LEH.